Forward Model Emulation for NASA’s Microwave Limb Sounder

• Activity Number: 271 - Statistical Modeling and Uncertainty Quantification for Atmospheric Remote Sensing Retrievals
• Type: Topic Contributed
• Date/Time: Tuesday, August 4, 2020 : 1:00 PM to 2:50 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): Margaret Johnson* and Joaquim Teixeira and Nathaniel Livesey and Amy Braverman
• Companies: Jet Propulsion Laboratory and Jet Propulsion Laboratory and Jet Propulsion Laboratory and Jet Propulsion Laboratory, California Institute of Technology
• Keywords: uncertainty quantification; Gaussian process emulation; atmospheric retrievals

Abstract

NASA's Microwave Limb Sounder (MLS) has been collecting data on the chemistry and dynamics of the upper troposphere, stratosphere, and mesosphere since its launch aboard EOS-Aura in July 2004. MLS scans the “forward” limb, and ground data processing software retrieves vertical profiles of temperature, water vapor, and other constituents, along the Aura orbit track. The current MLS retrieval algorithm utilizes a computationally expensive forward model to relate vertical profiles of constituents to radiance measurements. In this talk, we present a Gaussian process-based emulator of the MLS forward model for use in simulation-based uncertainty quantification and to facilitate computationally efficient approximate retrievals. The methodology incorporates dimension reduction through multivariate functional principal component analysis in the radiance space and kernel-based sufficient dimension reduction in the state space. A likelihood approximation is used to estimate independent Gaussian processes in the reduced space using a large sample of historical MLS retrievals. Performance of the emulator is compared to that of the MLS full and linearized forward models.